Tensor Radial Lengths for Mammographic Image Enhancement

Abstract

Proper enhancement of mammographic images in order to reveal diagnostically critical information is a very important task in everyday clinical practice. As the volume of screening mammograms increases so does the importance of algorithms that can reveal tumors or other kind of lesions. In this paper a mammogram enhancement method, inspired from the concepts of tensor image representation and tensor scale, is presented. In particular a number of tensor radial lengths is defined at each image location and their mean value is then subtracted from the original image. The proposed method was tested on a dataset of 192 images containing mass lesions taken from the Digital Database for Screening Mammography providing quite promising results. Furthermore the enhancement performance of the proposed method was compared with the enhancement performance of Contrast Limited Adaptive Histogram Equalization, Histogram Equalization and Unsharp Masking methods and clearly outperformed them.